0
$\begingroup$

Is there an elementary way to see that a product of an infinite number of non-units in a (noncommutative) ring is always $0$? Here, by non-units I mean elements that are not left-invertible.

This statement should be equivalent to every principal left ideal being Noetherian. Since the latter is true for any ring, the former must also be true, but I couldn't find an elementary proof.

$\endgroup$
3
  • 3
    $\begingroup$ Infinite products only make sense in topological rings. This is because a binary pruducf can only be extended to finite products, algebraically. $\endgroup$ Oct 30, 2022 at 20:36
  • 1
    $\begingroup$ It is not true in $\mathbb Z$ with the usual topology. But it is true if you add the condition "If the infinite product of non-units converges, it converges to $0.$" Not sure if that would be true in general, though. $\endgroup$ Oct 30, 2022 at 20:41
  • 1
    $\begingroup$ It seems unlikely to be true. For example, in the ring $R=\mathbb Z[1/2]$ with the usual topology, I think we can fairly easily find non-units whose product is $1.$ $\endgroup$ Oct 31, 2022 at 0:04

1 Answer 1

2
$\begingroup$

It is not true, even if you have a topological ring, and even if you assume the infinite product exists.

For example, the ring: $$\mathbb Z[1/2]=\left\{\frac a{2^k}\mid a,k\in\mathbb Z, k\geq 0\right\}$$ with the topology inherited from the rational numbers.

The units are $\pm 2^j$ for any $j\in\mathbb Z.$

Then the product: $$\frac{3}{4}\cdot \frac{5}{4}\cdot \frac{17}{16}\cdots \frac{2^{2^k}+1}{2^{2^k}}\cdots$$

converges to $1,$ but all of the terms are non-units. (The first term, $\frac34,$ does not match the general pattern, but the other terms do.)

A partial product, up to the term for $k,$ can be shown to be $$\frac{2^{2^{k+1}}-1}{2^{2^{k+1}}}$$ by induction.

$\endgroup$
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .